To provide high bit rate transmission over existing telephone subscriber loops, various modem technologies have been proposed. One of the promising solutions is the Asymmetric Digital Subscriber Loop (ADSL) technology that can provide up to 6.144 Mb/s transmission from the central office to a subscriber (downstream) and up to 640 kb/s transmission from the subscriber to the central office (upstream).
As the DSL technology rapidly advances, there is a strong need for the carrier (i.e. phone companies) to provide cost-effective, end-to-end, and high-speed interconnection. However, as explained below, there are many complex issues arising at both the upstream and downstream sites that make it difficult to develop cost-effective and easy-to-install and use solutions.
First, because of the earlier PCM (pulse code modulation) design where analog voice is digitized at a rate of 64 kb/s, the digital telephone switches installed in the Public Switched Telephone Network (PSTN) currently provide only 64 kb/s end-to-end connections. For example, ISDN is a DSL technology that can provide end-to-end circuit switching at a rate of multiple 64 kb/s. Each 64 kb/s link in ISDN is called a B channel and users who want a circuit connection at a rate higher than 64 kb/s needs to use multiple 64 kb/s links at the same time. In this case, all source signals are digital (voice will be sampled to 64 kb/s at the user site) and transmitted over individual B channels. They can be switched by either a digital PSTN 115 or packet switching backbone network 120 as shown in FIG. 1A. In this case, ISDN has the following limitations: (1) The transmission rate over the ISDN line (i.e., from IDSN Network Terminal 110A to 110B) is fixed and cannot be expanded (e.g. basic rate ISDN is 128 kb/s and primary rate ISDN is 1536 kb/s). For high performance services such as video conferencing or graphic file transfers, this data rate is not useful and/or it takes too long in time to transfer. (2) Voice traffic is carried via 64 kb/s PCM or one B-channel. Compared to a typical basic rate access of 2 B-channels, voice connection consumes a large portion of the total bit rate. (3) The protocol for connection over packet-switching backbone network 120 is standardized and requires the other end to follow the same protocol. For ADSL access where transmission rates are in the order of Mb/s, use of a large number of B channels (i.e., multiple ISDN connections) is practically undesirable due to the cost of multiple fixed switched connections. Furthermore, even though the ADSL transmission rate is high, it may not require a constant transmission rate (as is provided by a typical ISDN direct switched connection) all the time for many practical applications such as Internet access.
To overcome the above problems, packet switching (in contrast to circuit switching) based solutions for xDSL such as ATM and Frame Relay have been proposed. The term "xDSL" generally refers to a superset of various digital subscriber loop technologies, including ADSL, HDSL, etc. In particular, WANs (as used herein, "WAN" refers to any packet-switching based network such as Frame Relay, ATM, or SMDS (Switched Megabit Data Service)) can provide packet-switched based connections at variable rates and have been proposed to support xDSL. An example of a WAN arrangement 180 is shown in FIG. 1B. In this arrangement, connections at a rate other than multiple 64 kb/s between two CPEs 130 and 132 can be established through WAN backbone data network 160 at a lower cost due to bandwidth sharing. Because they are very suitable for data transfer, these types of high-speed backbones have been widely used in LAN interconnections as well. However, they do not guarantee fixed transfer delay. Therefore, they are not suitable for time-sensitive services such as video conferencing. In addition, they require non-trivial network access setups. As a result, they are difficult for ordinary users to install and maintain. To terminate an xDSL line 125 and connect it to a WAN 160, a piece of equipment called DSLAM (DSL Access and Multiplexer) 140 is used. As shown in FIG. 1B, a DSLAM splits a subscriber loop 125 to a PSTN 150 for analog voice signals and the WAN 160 for data transmission. As shown, however, the above DSLAM based architecture has the following known limitations. (1) Data transmission always goes through the same backbone data network 160. It is desirable to be able to use the PSTN 150 for switching time-sensitive services. (2) As a result, this type of arrangement does not support end-to-end circuit switching other than Plain Old Telephone Services (POTs), and data communications using voice-band modems. This prohibits the use of the current suggested DSL "modem model" in which end-users at CPE 130 can "dial-up" any remote site 131, 132 with a compatible modem user model. Instead, users need to set up all the necessary network addresses for both the host and intermediate nodes. This can be troublesome for most end users and especially a problem when the network needs to be upgraded (ie. the network is no longer transparent to users). (3) It does not have the ability to split the data signals carried by the xDSL. 125 into two paths: one through the PSTN 150 and one through the WAN 160. The access to the WAN 160 needs to support the maximum xDSL rate. If not, the high-speed transmission over the DSL becomes wasted. On the other hand, the access cost to the WAN for this type of data rate can be expensive. This poses a challenging problem for the carriers to price xDSL access.
Furthermore, the cost of DSL, codecs and access equipment are currently much higher than that of voice-band modems. Therefore, even though the speed is much higher than the current 33.6 kb/s or 56 kb/s, most end-users will not afford to upgrade this new technology. A lower cost alternative is thus desirable that users can spend less initially for a lower speed and upgrade it at a later time as demands increase.
In addition to the equipment cost, xDSL users will have to spend much more for the access to a high-speed backbone network. In contrast to the current case where modem users do not need to pay any additional cost, this poses another barrier for adoption of xDSL technology. Users who subscribe to Frame Relay or T1 access typically need to spend $1,000 or even more every month, a figure which is beyond the means of the majority of potential users of such technology.
A critical need, therefore, exists for a solution that minimizes accesses charges while at the same time allowing carriers to enjoy a reasonable commercial return on their investments in higher end equipment to provide ADSL services. To address this need, a forward compatible and expandable xDSL modem or so-called "SAM" (scaleable ADSL modem) has been proposed as a low cost solution at the end-user side in pending U.S. application Ser. No. 08/884,995 filed Jun. 30, 1997 entitled "Rate Adaptable Modem With Forward Compatible and Expandable Functionality and Method of Operation," also assigned to the present assignee. The invention of that SAM disclosure makes it possible for downstream users to avoid the cost associated with an expensive ADSL modem when they do not need full ADSL transmission rate. In an analogous fashion, it would be attractive and advantageous to extend some of the principles of the above SAM disclosures to the upstream sites. In other words, the central office should be able to effectuate an end-to-end architecture that: (i) permits users to only pay carriers a fee necessary to procure a particular desired target data rate (which may be only a fractional portion of a full ADSL data link); (ii) allows users to establish a particular kind of data link (real-time or delayed); (iii) allows users to accept the lowest cost per unit of bandwidth; or (iv) facilitates a data route which is more suited to particular user's connection model.